akanametov / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors

Official YOLOv7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors

Performance

Model	Test Size	AP50test	batch 1 fps	batch 32 average time
YOLOv7-Tiny	640	98.7%	417 fps	2.4 ms

Installation

Docker environment (recommended)

Expand

# create the docker container, you can change the share memory size if you have more.
nvidia-docker run --name yolov7 -it -v your_coco_path/:/coco/ -v your_code_path/:/yolov7 --shm-size=64g nvcr.io/nvidia/pytorch:21.08-py3

# apt install required packages
apt update
apt install -y zip htop screen libgl1-mesa-glx

# pip install required packages
pip install seaborn thop

# go to code folder
cd /yolov7

Testing

yolov7-tiny-builder.pt

python test.py --data data/builder.yaml --img 640 --batch 32 --conf 0.001 --iou 0.5 --device 0 --weights yolov7-tiny-builder.pt --name yolov7_tiny_builder_val

You will get the results:

   Class      Images      Labels           P           R      [email protected]  [email protected]:.95
     all         200        1376       0.985       0.969       0.987       0.679
    vest         200         486        0.98       0.986       0.997       0.707
  helmet         200         404       0.992       0.928       0.968       0.573
  person         200         486       0.984       0.994       0.997       0.757

Inference

On image:

python detect.py --weights yolov7-tiny-builder.pt --conf 0.25 --img-size 640 --source inference/images/builder/1779.jpg

Results

PR curve:

Losses and mAP:

Confusion matrix:

Training

Data preparation

bash scripts/get_dataset.sh

• Download dataset and pretrained yolov7-tiny.pt

Single GPU training

# train tiny model
python train.py --epochs 100 --workers 4 --device 0 --batch-size 32 --data data/builder.yaml --img 640 640 --cfg cfg/training/yolov7-tiny-builder.yaml --weights 'yolov7-tiny.pt' --name yolov7_tiny_builder --hyp data/hyp.scratch.tiny.yaml

Transfer learning

yolov7-tiny.pt

Citation

@article{wang2022yolov7,
  title={{YOLOv7}: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors},
  author={Wang, Chien-Yao and Bochkovskiy, Alexey and Liao, Hong-Yuan Mark},
  journal={arXiv preprint arXiv:2207.02696},
  year={2022}
}

Acknowledgements

Expand

• https://github.com/AlexeyAB/darknet
• https://github.com/WongKinYiu/yolor
• https://github.com/WongKinYiu/PyTorch_YOLOv4
• https://github.com/WongKinYiu/ScaledYOLOv4
• https://github.com/Megvii-BaseDetection/YOLOX
• https://github.com/ultralytics/yolov3
• https://github.com/ultralytics/yolov5
• https://github.com/DingXiaoH/RepVGG
• https://github.com/JUGGHM/OREPA_CVPR2022
• https://github.com/TexasInstruments/edgeai-yolov5/tree/yolo-pose